CTCF-mediated transcriptional regulation through cell type … · 2017. 1. 29. · CTCF-mediated...

CTCF-mediated transcriptional regulation through cell

type-specific chromosome organization in the β-globin

locus.

Ivan Junier, Ryan K Dale, Chunhui Hou, Francois Kepes, Ann Dean

To cite this version:

Ivan Junier, Ryan K Dale, Chunhui Hou, Francois Kepes, Ann Dean. CTCF-mediated tran-scriptional regulation through cell type-specific chromosome organization in the β-globin locus..Nucleic Acids Research, Oxford University Press (OUP): Policy C - Option B, 2012, epub aheadof print. <10.1093/nar/gks536>. <hal-00732516>

HAL Id: hal-00732516

https://hal.archives-ouvertes.fr/hal-00732516

Submitted on 14 Sep 2012

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinee au depot et a la diffusion de documentsscientifiques de niveau recherche, publies ou non,emanant des etablissements d’enseignement et derecherche francais ou etrangers, des laboratoirespublics ou prives.

https://hal.archives-ouvertes.fr

https://hal.archives-ouvertes.fr/hal-00732516

CTCF-mediated transcriptional regulation throughcell type-specific chromosome organization in theb-globin locusIvan Junier1,2,3,*, Ryan K. Dale4, Chunhui Hou4, Francois Kepes1,* and Ann Dean4,*

1Epigenomics Project and institute of Systems and Synthetic Biology, Genopole!, CNRS, University of Evry,5 rue Henri Desbrueres, Evry F-91030, 2Institute of Complex Systems, Paris, France, 3Centre for GenomicRegulation, C/ dr. Aiguader, 88, 08003 Barcelona, Spain and 4Laboratory of Cellular and DevelopmentalBiology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health,Bethesda, MD 20892, USA

Received February 24, 2012; Revised May 4, 2012; Accepted May 11, 2012

ABSTRACT

The principles underlying the architectural land-scape of chromatin beyond the nucleosome level inliving cells remains largely unknown despite its po-tential to play a role in mammalian gene regulation.We investigated the three-dimensional folding of a1 Mbp region of human chromosome 11 containingthe b-globin genes by integrating looping inter-actions of the CCCTC-binding insulator proteinCTCF determined comprehensively by chromosomeconformation capture (3C) into a polymer model ofchromatin. We find that CTCF-mediated cell type-specific interactions in erythroid cells are organizedto favor contacts known to occur in vivo between theb-globin locus control region (LCR) and genes.In these cells, the modeled b-globin domain foldsinto a globule with the LCR and the active globingenes on the periphery. In contrast, in non-erythroidcells, the globule is less compact with few butdominant CTCF interactions driving the genes awayfrom the LCR. This leads to a decrease in contactfrequencies that can exceed 1000-fold dependingon the stiffness of the chromatin and the exactposition of the genes. Our findings show that an en-semble of CTCF contacts functionally affects spatialdistances between control elements and targetgenes contributing to chromosomal organizationrequired for transcription.

INTRODUCTION

Chromatin interactions that form loops between enhan-cers and target promoters over large linear distances arecommon in metazoans (1–3). The specificity of such inter-actions is thought to be influenced by DNA–proteincomplexes called insulators, also through the formationof looped domains (4). How enhancer–promoter proxim-ity is established and how topological domains contributeto transcription activation is not known. Moreover,recent technological advances have revealed genome-widenetworks of intra- and inter-chromosomal sites of contact(5–7). Thus, a major question that arises is to what extentnuclear proximities between sites in the genome reflectspecific long-range interactions with consequences forthe transcriptional activity of underlying sequences.

The b-globin locus is flanked by two sites, HS5 and30HS1, that are occupied by the insulator protein CTCF.Chromosome conformation capture (3C) experimentshave shown looping together of these sites in erythroidprecursor cells before the globin genes are transcribed(8). Encompassed in this loop are the b-globin family ofgenes, 50 embryonic e, fetal Ag and Gg and 30 adult b andthe locus control region (LCR) that activates them sequen-tially in erythroid cells (9). Globin genes are subsequentlyfound in proximity to the LCR as they are activatedduring development (10). These looping interactionsrequire erythroid transcription factors GATA-1, FOG-1and EKLF, and the ubiquitous nuclear factor Ldb1, likelyas participants in multimeric protein complexes bridgingthe LCR and genes (11–13). In addition, the chromatin

*To whom correspondence should be addressed. Tel: +1 301 496 6068; Fax: +1 301 496 6068; Email: [email protected] may also be addressed to Francois Kepes. Tel: +33 169474431; Fax: +33 169474437;Email: [email protected] may also be addressed to Ivan Junier. Tel: +34 93 3160166; Fax: +34 93 3969983; Email: [email protected] address:Chunhui Hou, Department of Biology, Emory University, Atlanta, GA, USA.

7718–7727 Nucleic Acids Research, 2012, Vol. 40, No. 16 Published online 16 June 2012doi:10.1093/nar/gks536

Published by Oxford University Press 2012.This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

by Ivan Junier on September 12, 2012

http://nar.oxfordjournals.org/D

ownloaded from

remodeler Brg-1 is required at an early time in the loopingprocess (14).

CTCF has been implicated in the organization of indi-vidual chromosomes and chromosome territories (15), anda role for CTCF in both organizational and transcrip-tional regulation has been suggested by the correlationof CTCF-mediated loops genome wide with lamin-associated domains demarcating regions of differingepigenetic marks (6). However, a functional role forCTCF site interactions has been probed in only limitedinstances (16). For example, the CTCF-dependent im-printing control region (ICR) in the Igf2/H19 locusforms an enhancer blocking loop with a site upstream ofIgf2 on maternal chromosomes, isolating the gene fromenhancers it shares with H19 (17,18). Loss of CTCF orits sites results in loss of the loop and activation of Igf2. Incontrast, individual deletion or disruption of b-globin HS5or 30HS1 CTCF sites did not affect b-globin transcriptionin mouse erythroid cells (19–22).

Recent integrative numerical frameworks have allowedthe generation of spatial conformations of chromosomesin silico that are compatible with 3C-like high-throughputdata (23–28). More causal approaches have been used tostudy the mechanisms that may be responsible for chr-omosome organization by modeling chromatin as apolymer (7,29–34). In a theoretical analysis using such apolymer model, Mukhopadhyay et al. suggested that longrange interactions could have a crucial impact on thecontact properties of genes and their enhancers (35).How chromatin configurations may influence gene expres-sion remains unclear.

Here, we investigate whether regional long-range CTCFcontacts contribute to transcriptional control at theb-globin locus. To this end, we constructed chromosomemodels that satisfy the constraints of long-range CTCFdependent interactions over 1Mb of human chromosome11 encompassing the b-globin locus (36). The modelsconsist of a single, self-avoiding polymer chain alongwhich specific sites (the CTCF sites here) are able tointeract (30). Using 3C data as input, simulations of thebest-fit models make it possible to quantify the spatialproximity between the globin genes and their LCR inerythroid K562 cells and non-erythroid 293T cells. Theresults show that the set of regional CTCF site inter-actions drive the LCR and globin genes closer togetherin expressing cells than in silent cells and nucleate aglobule with the contacting LCR/globin gene pairdeployed away from the center.

MATERIALS AND METHODS

Modeling the chromatin around the b-globin locus: aself-avoiding WLC with designated interacting sites

We simulated 1Mbp long chromosome segments corres-ponding to the set of interacting CTCF sites in each celltype (Figure 1), using a polymer model of the chromatinfiber (37) (Supplementary Figure S1). In this worm-likechain (WLC) model, the chromatin fiber behaves as aflexible chain (37–39). We further excluded the possibilityof the chromatin fiber overlapping itself (self-avoidance

effect) by specifying the chain diameter to be 30 nm. Thebase-pair density along the fiber was fixed at 150 bp/nm(37). In order to prevent the formation of knots, simulatedchromosomes were circularized by adding a 0.2Mbp longfiber free of CTCF sites (not represented in the figures),which is long enough (1.3 mm) to prevent looping of theends of the 1Mbp chromosome. This led to a final1.2Mbp chromosome segment, which was confined to aspherical volume of 1 mm diameter.Along the chromatin, designated sites akin to CTCF

sites interact with each other according to a harmonicpotential VðxÞ ¼ k

2 x2 whose value is lowest for a spatial

distance between the sites, x, that is equal to zero. For anygiven pair of known interacting sites, the spatial distance,x, separating them varies according to the chromosomeconformation and the larger the intensity, k, of theirpotential, the more effective the potential V(x), that is,the more frequently the sites interact. Unlike modelswhere interactions are short-range (30), long-range poten-tials such as harmonic ones prevent interacting sites frombeing spatially distant from each other. This allowed usto comprehensively study the statistical properties ofthe models by running numerous short (i.e. <1 day)simulations.

Fitting the polymer models: an iterative procedure

3C experiments quantify interaction frequency betweentwo sites. For input to the models, these values were con-verted to potential intensities. To this end, the potentialintensities for a given set of 3C contact frequencies wereiteratively updated to obtain contact frequencies in silicoas close as possible to contact frequencies in vivo(Supplementary Material). In addition, we do notconsider interactions between sites that are in contactwith a common third site in order to avoid non-lineareffects on the uniqueness of the solution of the iterativealgorithm (Supplementary Figure S2).

Computing statistical properties of best-fit polymer models

Given a best-fit polymer model, we used distributedcomputer facilities in order to sample an equilibrium setof conformations as large as possible. An equivalent com-putation time of 3100 days ($8.5 years) was run on theEuropean Grid Infrastructure (EGI) using two virtual or-ganizations (VO): biomedical computing (biomed) andcomplex systems computing (vo.iscpif.fr, see http://www.iscpif.fr/tiki-index.php?page=csgrid). The use of thedistributed facilities was accomplished through theOpenMOLE project (http://www.openmole.org/) (40).For a more detailed description of the methods, seeSupplementary Data.

RESULTS

CTCF site long-range interactions surroundingthe b-globin locus are primarily cell type specific

In a previous study, we used chromatin immunopre-cipitation combined with microarray analysis (ChIP-chip) and ChIP and qPCR to survey and validate CTCF

Nucleic Acids Research, 2012, Vol. 40, No. 16 7719



ownloaded from

sites up- and down-stream of the b-globin locus on humanchromosome 11 in K562 cells, where the fetal g-globingene is transcribed and in 293T cells where the globinlocus is silent (36). A wealth of structural information isavailable for K562 cells through the ENCODE consor-tium and these cells demonstrate the appropriateLCR-g-globin chromatin looping while 293T cells do not(41) (Kiefer,C.M. and Dean,A., unpublished data). Allsites analyzed were similarly occupied by CTCF in thetwo cell types and all were co-occupied by cohesincomplex members Rad21, SMC1 and SMC3. We thenused 3C to determine, in these two cell types and inK562 cells in which CTCF had been reduced usingRNAi (K562kd), the frequencies of interaction amongall occupied CTCF sites surrounding the globin locusover a chromosomal range of 1.4Mbp. 3C quantifies theinteraction frequency between two sites in chromatin afterformaldehyde cross-linking of protein and DNA, restric-tion enzyme digestion, re-ligation and detection of novelsequence junctions by qPCR (42).A representation of the interactions previously observed

(36) is presented in Figure 1. Each black dot represents anoccupied CTCF site. Each parabola represents a pairwiseCTCF site interaction where the height of the curve indi-cates the relative frequency of interaction on a log scale.Also represented are the decreased interactions betweenthese sites when CTCF is knocked down in K562 cellsby RNAi, reducing g-globin transcription and allowingincursion into the locus of the repressing histone modifi-cation H3k9me2. Although CTCF was bound at nearlyidentical locations throughout the locus in the differentcell types, cell type-specific patterns of long-range

interactions distinguish K562 cells (red) from 293T cells(blue) with 60% of interactions being cell type-specific.Interestingly, the loop between b-globin 30HS1 and HS5(C15 and C16, respectively) was observed in both celltypes. In K562kd cells, all interaction frequencies weresignificantly reduced compared to CTCF-replete cellsbut the pattern of interactions was unchanged. Thesedata raise the possibility that CTCF organizes the chr-omosomal domain structure surrounding the globinlocus to influence transcription.

Modeling cell type-specific interactions betweenCTCF sites

To address the issue of whether differential cell type-specific CTCF contacts affect the conformation of theglobin region, a polymer model of the 30 nm chromatinfiber was designed for each of the three cell types underinvestigation: K562, K562kd and 293T. Disregarding itsinternal structure, chromatin behaves as a flexible chainthat self-avoids. We used the simplest continuous model,known as the WLC, which includes flexibility properties ofa single chain (37–39) and considers the self-avoidingnature of a 30 nm thick fiber. Each WLC model representsa 1Mbp region of human chromosome 11 containing theinteracting CTCF sites and the b-globin locus. Interactionforces were determined by iterating the DNA-foldingalgorithm until the frequency of contacts between thesites corresponded to the experimentally measuredCTCF-dependent chromatin contacts (Figure 1) (36).Importantly, CTCF interaction frequencies were the soleinput to the polymer models and the inputs did not includethe LCR or any gene promoter sites. This allowed us to

Figure 1. CTCF-based chromatin loops differ between K562 cells and 293T cells. CTCF site interactions from Hou et al. (36) are depicted in aschematic diagram of the globin locus. Pairwise interaction frequencies detected by 3C are shown by arcs for K562 (red), K562 CTCF RNAiknockdown (green) and 293T (blue) cells. Black dots below the x-axis indicate CTCF-occupied sites confirmed by qPCR used in this analysis. TheLCR is indicated by a gray vertical line. RNA seq data from the ENCODE Consortium (CalTech data set from Wold and Myers groups) for K562cells (red track) and nine other non-erythroid cell types combined (multicolor track) are shown as normalized read density (reads per million, RPM)on a ln(x+1) scale. RefSeq gene models for the locus are shown, including globin genes (orange), odorant receptor genes (blue) and other genes(gray). All coordinates and data are for the hg19 assembly.

7720 Nucleic Acids Research, 2012, Vol. 40, No. 16



ownloaded from

make independent in silico measurements of LCR–promoter proximity during the model runs using theseproximities as a ‘read-out’ of the potential transcriptionalstate of the locus.

3C experiments measure interactions in a population ofcells. We assume that the measured population-averagedfrequencies are analogous to time-averaged frequencies atequilibrium in the models—that is, averaging over popu-lations is equivalent to averaging over time. Two param-eters of the model are adjustable, namely, the strength ofinteraction between the CTCF sites (interaction potential)and the persistence length, which describes the flexibil-ity status of the chromatin fiber—the larger the persistencelength, the less flexible the chromatin fiber (Sup-plementary Figure S1). Since the persistence length isnot precisely known in vivo, we studied three values(100, 200 and 300 nm) typical of those reported in the lit-erature (38). We then iteratively and independentlyadjusted intensities of the interaction potentials so thatequilibrium contact frequencies between CTCF sites bestmatched experimentally measured frequencies (36).These best-fit interaction potentials were subsequentlyused in a benchmark polymer model to compute statisticalproperties, such as equilibrium distributions of the spatialdistances between the globin genes and the LCR andtypical chromatin conformations.

The pattern of intensities obtained for the interactionpotentials in K562 and K562kd cells were comparable(Supplementary Table S1), with intensities reduced inK562kd cells compared to K562 cells. This indicates thatinteractions are based on a similar but weakened mechan-ism, in accordance with the incomplete knockdown byRNAi which allows some CTCF protein molecules toremain associated with chromatin (36). In contrast,intensities obtained for the potentials in 293T cells wereglobally much lower than those in K562 cells, consistentwith the overall lower frequency of 3C interactionsobserved in 293T cells. More specifically, two categoriesof interactions of qualitatively different natures can bedistinguished in 293T cells. The first category consists ofthe pairs C-08/C-20 and C-20/C-21 (Figure 1) for whichthe interaction intensities were relatively strong. Thesecond category contains all the other pairs of interactingsites for which the intensities were at least two orders ofmagnitude lower (Supplementary Table S1).

Sites within the globin locus domain are in closerproximity to the LCR in erythroid cells than innon-erythroid cells

Next, we sought to tie together the modeled interactions inthese cells with known expression patterns. Since theprimary difference between the cell types is the expressionof globin genes, and interaction with the LCR is requiredfor transcribing these genes, we chose to use the proximityof the LCR to a series of targets in the locus as a proxy fortranscriptional activity. The primary set of targets con-sisted of the transcription start site (TSS) of each of theglobin genes in the locus. In addition, we included the TSSfor the b-globin pseudogene, two intergenic sites and along-distance site downstream from b-globin (Figure 1).

The distributions of 3-D spatial distances between theLCR and each gene were measured for each cell type overthe course of 100 simulations of the corresponding best-fitmodel. Frequencies and cumulative frequencies show thetime fraction two sites spend at a given distance or at adistance lower than a given distance, respectively. Theseare displayed in Figure 2 for a representative target, theGg-globin gene, for all simulations (thin curves) and forthe corresponding averages (thick curves). Figure 2A dem-onstrates a clear shift of the distributions toward largedistances for 293T cells compared to K562 cells.Looking across all targets (Supplementary Figure S3), asgenomic distance from the LCR increases, so does thethree-dimensional spatial distance. This effect is strongerin 293T cells than in K562 cells so that at the largestgenomic distances, there is the greatest differencebetween cell types. Cumulative frequencies, shown on alog scale, reveal a tendency for the globin genes to becloser to the LCR in K562 than in K562kd (Figure 2Band Supplementary Figure S3). Importantly, the results donot qualitatively depend on microscopic parameters (thepersistence length) of the chromatin fiber. Taken together,these results show that in K562 cells, regional CTCF sitesinteract in such a way that globin genes are closer to theLCR than in 293T cells, and to a lesser extent, than inK562kd cells.

Effects of chromatin features and genomic organizationon LCR contacts

We next studied how the physical features of the chroma-tin fiber affect the capacity of globin genes to makecontact with the LCR. We compared results obtainedfor three levels of stiffness of the chromatin fiber typicalof those reported in vivo, which are defined by persistencelengths equal to 100, 200 and 300 nm, respectively(Figure 3). Since sites are likely to interact throughprotein complexes (e.g. transcription factors, RNA poly-merase), we chose a value of 10 nm to represent the spacetaken up by these complexes, and defined an interactionevent to exist when the LCR was within 10 nm of a targetsite, i.e. when there was %40 nm between the centers of thetwo 30-nm chromatin fibers. We then quantified the fre-quency of LCR–target interaction events across the locus.The frequency of an interaction event can be shown on

a negative log scale so that higher values indicate greaterisolation of the LCR from targets. This analysis highlightstwo important points (Figure 3). First, the stiffer the chro-matin fiber, the larger the difference between 293T celland K562 cell contact frequencies. For example, the iso-lation of Gg from the LCR is $20-fold larger in 293T thanin K562 at a persistence length of 100 nm; it is at least1000-fold larger at a persistence length of 300 nm, whichis within the range of the expected persistence lengthsin vivo (38). Second, the smaller the genomic distance tothe LCR, the larger the difference in contact frequenciesbetween K562 and K562kd cells. For example, at a per-sistence length equal to 200 nm, LCR contacts withb (genomic distance=53 kbp) are only $1.5-fold lessfrequent in K562kd cells than in K562 cells; contactswith g (genomic distance=26 kbp) are $4.5-fold less




ownloaded from

frequent. The observation that the Gg-globin gene is moreisolated from the LCR in K562kd cells than inCTCF-replete K562 cells is consistent with the reductionof Gg-globin expression by RNAi-mediated CTCF knockdown (36).

Spatial conformations of the globin locus during activetranscription of the c-globin genes

Compared to 293T cells, CTCF interactions in K562 cellsprovide an overall modeled conformation of the globinlocus that favors contact between the LCR and globingenes (Figure 3). We ran 100 additional simulations of aK562 polymer model to study the topology of the chro-matin conformations that involved LCR–globin geneinteractions. For each simulation, snapshots were storedof the first configuration that was reached when the LCRwas in contact with a globin gene of interest. Figure 4Ashows one such configuration for Gg (see SupplementaryFigure S4 for other genes).Like the 3C contact maps to which interaction poten-

tials were fitted (Figure 1), the snapshots reveal conform-ations that are composed of one globule, where many

contacts are present (blue sites), plus one loop whereonly two sites interact (red sites). These two structurestend to repel each other, due to volume exclusion effects(43). Interestingly, the contacts between the LCR andGg gene (green sites) occur most often on the peripheryof the globule. We quantified this with a radial distribu-tion function, which shows the probability that a site isfound at a certain distance away from the center of massof the globule.

Whereas the LCR does not have, on average, apreferred location with respect to the center of mass ofthe globule (compare red and black dotted curves inFigure 4B), interaction events between Gg and the LCRtend to happen farther out from the center of the globulethan is typical for the locus as a whole (Figure 4B,

Figure 3. Interaction frequency as a function of the genomic distanceto the LCR. The panels show the co-logarithm of the contact frequencybetween the LCR and b-globin genes, the intergenic sites (i1, i2) andthe downstream site (ds) for three persistence lengths. Higher valuesindicate increased isolation of LCR from target. A contact is definedwhenever the distance between the sites along the polymer and theLCR is lower than 40 nm. This figure shows that the stiffer the chro-matin, the lower the contact frequency in 293T cells. Note that due tonumerical limitations, frequencies lower than 10&6 cannot be detected:they are reported to be equal to 10&6 (e.g., for Gg at a persistencelength of 300 nm).

Figure 2. Distribution of distances between Gg and the LCR using abest-fit polymer model of the b-globin locus. The curves show the dis-tributions of the spatial distances (in nm) separating the LCR from Ggin the three cell types K562, K562kd and 293T, for a chromatin fiberwith a 200 nm persistence length. Thin dotted curves are distributionsobtained from 100 different runs, which were used to compute theaverage curves (thick solid curves). (A) The y-axis reports the timefraction the two sites spend at the distance reported along the x-axis.(B) The y-axis reports the time fraction the two sites spend at a distancelower than that reported along the x-axis. Note that at distances lowerthan 100 nm, the cumulative frequency is 10–100 times larger in K562cells than in 293T cells.




ownloaded from

blue; note that the blue line is less smooth since LCR-Gginteractions only occur during a subset of all time stepsrepresented by the red and black curves). Extending thisanalysis to Ag and b indicates that all globin genes, butparticularly g-globin genes, tend to be located more per-ipherally to the globule regardless of LCR contact(Figure 4C). In contrast, in 293T cells, where the globuleis less compact, no preferential location is observed forany of the locus sites of interest (Figure 4D). Thesefindings suggest that, in addition to favoring contactswith the LCR, the CTCF-driven globule in K562cells tends to displace the genes to be activated, i.e. theg-globin genes here, away from the surroundingchromatin.

Dominant CTCF interactions and stiff chromatin preventcontacts between the LCR and globin genes in 293T cells

The interaction potentials observed in 293T cells can bedivided into two categories based on strength (Sup-plementary Table S1). The strongest potentials arebetween C-08 and C-20 and between C-20 and C-21.

A polymer model where these interactions alone arepresent leads to a reduction of the tendency for globingenes to be spatially close to the LCR when the chromatinfiber is stiff (Supplementary Figure S5). To investigate theinfluence of these interactions, in particular whether thestrongest interactions found in 293T cells are sufficient todecrease LCR–gene interactions compared to K562 cells,we used two additional models: one where only the twostrongly interacting sites are present (ignoring all otherinteractions measured by 3C in 293T cells) and anotherusing chromatin with no interacting sites. Since the inter-action events we defined earlier (40 nm between chromatinfiber centers) do not always occur in 293T cells as they doin K562 cells, we used the minimal distance obtained in100 simulations as an alternative metric to representLCR–target proximities.The model with no interacting sites serves as a baseline

(red lines, Figure 5). One might hypothesize thatintroducing any interacting sites in this locus wouldbring the LCR closer to targets on average. However,interestingly, the model with just two pairs of strongly

Figure 4. Chromatin conformations favoring contacts between the b-globin genes and LCR in K562 cells. (A) Typical conformation of the 1Mbpregions around the b-globin locus during a contact between LCR (green+star) and Gg (green). Blue sites: CTCF sites that form a connected networkof interaction (Supplementary Figure S1). Darkest blue sites: CTCF sites that surround the b-globin locus. Red sites: the isolated interaction betweenC-08 and C-10. The conformation can be divided into a loop (stabilized by the red sites) and a compact globule (dashed orange ellipse) encompassingthe region from C-03 to C-10. (B) Spatial location of the contact: using 1000 equilibrium simulations of the same best-fit polymer as in A, we report(i) the radial mass distribution of the compact globule, i.e. the average probability density for the location of the C-03 to C-10 region with respect tothe globule center of mass; (ii) the radial distribution of Gg and LCR during contacts and (iii) the radial distribution of the LCR (no matter theposition of Gg). One can see that the Gg/LCR contacts tend to occur away from the globule center. (C) Spatial location of the globin genes in K562(obtained from 100 simulations of the best-fit polymer). Genes tend to be located away from the center regardless of LCR contact. Large distancesare particularly enhanced in the case of the g genes. (D) Same as in C but for 293T cells. No particular location can be observed for any of the genes.




ownloaded from

interacting sites results in a substantial isolation of theLCR from the targets over most of the locus comparedto the model with no interacting sites (Figure 5, comparepurple and red lines). This strong isolating effect decreaseswhen the rest of the 293T interaction potentials areincluded (Figure 5, blue lines), even though the shortminimal distances reported here remain very infrequent(Figure 3). These additional interactions tend tore-establish contacts of a subset of the globin genes, e.g.b (Figure 5, upper panel). The isolating effect of the twopairs of strongly interacting sites suggests that they mayplay an important role in cell type-specific chromosomearchitecture surrounding the b-globin locus.

DISCUSSION

Above the megabase scale, Hi-C studies have revealed thatcontact maps of individual human chromosomes can bedivided into two or three classes of regions, with one classthat is more transcriptionally active than the other(s)(7,44). Novel conformation capture techniques have

further revealed that inter-chromosomal contacts notonly correlate with transcriptional activity but also withCTCF-bound regions (26,44). Between the megabase scaleand the nucleosome level of chromatin folding little isknown about chromatin organization. In this study, weuse a thermodynamic framework based on a WLCmodel of chromatin where designated sites along theDNA, here CTCF sites over 1Mbp around the b-globinlocus, are able to form pairwise interactions (30,36). Theresults indicate that (i) proximity within the globin locusbetween the LCR and genes and (ii) displacement of theactive LCR/globin pair away from surrounding chromatinare both facilitated by overall erythroid cell type-specificCTCF chromosome organization. Thus, providing athermodynamic topological model with experimentalmeasurements of interaction frequencies of CTCF sitesalone predicts a chromosome conformation pattern con-sistent with the known biology of the human b-globinlocus.

Our analysis uses long-range interaction potentials toefficiently sample conformations that are consistent withthe CTCF contact map. The main advantage of this choiceis that it allows a comprehensive statistical analysis ofchromosome folding. The disadvantage is that real inter-actions have a shorter range in the cell, and the choiceof long-range interactions may alter the statisticalproperties of the simulated chromosome conformations.Nevertheless, long-range interactions might mimic thereal situation in cells, where non-functional chromosomeconformations appear to be dis-favored by mechanismsthat remain elusive.

CTCF sites within 1Mbp of the b-globin locus wereoccupied in both K562 (active locus) and 293T (silentlocus) cells, however, their pairwise interactions were pri-marily cell type specific, suggesting that factors otherCTCF are important for the specificity of loop formation.In some loci, cohesin co-occupancy is the determinant ofCTCF insulator function, however, cohesin was presentat CTCF sites we studied in both K562 and 293T cells(45). Modifications to these proteins or recruitment ofadditional components may explain the difference inloop formation between the cell types. In addition,looping interactions beyond those we studied are likelyto impinge on chromosome organization in the b-globinregion, including those mediated by different proteinfactors (46,47). Nonetheless, our simulations indicatethat consideration of CTCF looping contacts alonepredicts a cell type-specific chromosome organization inwhich the probability of intra-loop LCR/globin genecontacts is favored for K562 cells and hindered in 293Tcells. Interestingly, numerous interacting sites around thea-globin locus were found to be bound by CTCF proteinand our previous data revealed that a-globin transcriptionwas reduced upon CTCF knock down (36,48). Thisfurther reinforces the idea of an important role ofregional CTCF sites in the interplay between the architec-ture of chromosomes and their function.

The modeled distances from the LCR to globin genes asa whole are smaller in K562 cells, where the locus is active,than in 293T cells where the locus is inactive. Moreover,reduction of CTCF in K562 cells using RNAi, which leads

Figure 5. Minimal distances between the LCR and the b-globin genes.Each point represents the minimal spatial distance obtained in oursimulations between a globin gene and the LCR at persistencelengths 200 and 300 nm. In K562 cells, the minimal distance is 30 nm,indicating that physical contacts are always possible. The situation isdifferent for 293T cells. In particular, when only the two strongestpotential pairs in 293T cells are considered (‘2pairs’), the minimaldistance can be as large as 80 nm (purple points). The red pointsshow the results where no interaction is considered. By taking intoaccount all the interactions involved in 293T cells, one can observefor persistence lengths of the chromatin around 300 nm a completeisolation specific to Gg and points linearly closer to the LCR.




ownloaded from

to a reduction of g-globin transcription (36), increasesthese modeled distances compared to cells replete withCTCF. Thus, CTCF site contacts in K562 cells mayprovide a constraint such that the collision dynamicsbetween globin genes and the LCR are the most appro-priate for LCR–globin gene interactions. We suggest thatthe relative closeness of the LCR to the coding regions ofthe locus modeled in K562 cells, compared to 293T cells,favors ‘sampling’ of contacts between the LCR and genesthat, if stabilized by protein–protein interactions, allowlooping and transcription activation. Our results indicategains in sampling time that can exceed 1000-fold, depend-ing on the stiffness of the chromatin and the position ofthe gene. In this view, the active K562 g genes would suc-cessfully establish LCR contact, whereas, the silent b genewould not, due to the modulation of required develop-mental stage-specific proteins or protein complexes. Thedata do not exclude the possibility that LCR/globin genecontact is the initial organizing event for the locus duringerythroid differentiation and that the CTCF-loopingensemble evolves to stabilize this interaction. Reductionof the LCR/globin gene interaction when CTCF isdecreased by RNAi shows the importance of the CTCF-mediated loop ensemble to maintain contact.

For transcription to occur, the LCR and gene to beactivated must be accessible to the transcription machin-ery. Our studies highlight in K562 cells a modeled prefer-ential exposure of the LCR and g genes to the nuclearmilieu at the surface of a CTCF-driven globule. This pos-itioning is reminiscent of extrusion from a chromosometerritory and might facilitate localization of the g-globin/LCR ensemble within nuclear foci of transcriptionto which globin genes are known to migrate to achievehigh levels of transcription (49–51). Interestingly,modeling of chromosome conformation capture carboncopy (5C) interactions in K562 cells suggested that thea-globin genes cluster together with neighboring activehousekeeping genes that surround them in the interior ofa globule that might correspond to a transcription factory(48). In contrast to the a-globin locus, the b-globin locus isembedded among odorant receptor genes that are silent inerythroid cells. Thus, the tendency for the active g-globingene and LCR to be on the periphery of a globule mightbe related to the difference in transcriptional statecompared to neighboring genes.

This study highlights the interplay of numerous CTCFsites within a chromosomal region that interact to formmultiple chromatin loops. The modeled conformation ofthe region reveals an ensemble of interactions (theglobule), whose overall distribution may have moresubtle biological implications (e.g. the propensity for Ggto be on the periphery) than suggested by consideration ofindividual loops. This may help to explain why individualdeletion or disruption of HS5 or 30HS1 (C-16 and C-15,respectively, in Figure 1) did not affect b-globin transcrip-tion in mouse erythroid cells (19–22). Likewise, in thisview, the insertion of an extra HS5 site (52) that createstwo new loops and limits access of the LCR to globingenes could be interpreted as introducing an imbalanceto the ensemble of interactions, similar to the two pairsof strong interacting CTCF sites in 293T cells that isolate

the LCR from the locus. These ideas suggest specific,testable biological hypotheses. For example, disruptingthe strongly interacting CTCF site pairs in 293T cellsshould reduce LCR isolation from globin genes,although such manipulation is unlikely to affect transcrip-tional activity due to the absence of erythroid transcrip-tion activators in these cells. Alternatively, incrementaldeletion of CTCF sites in the K562 computationalmodel could suggest which sites are most important forLCR/globin gene contact, and the predictions could thenbe tested experimentally.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online:Supplementary Table 1, Supplementary Figures 1–7 andSupplementary Material and Methods.

ACKNOWLEDGEMENTS

We thank Olivier Martin and Elissa Lei for helpfulcomments on the manuscript.

FUNDING

Intramural Program of the National Institute of Diabetesand Digestive and Kidney Diseases, National Institutesof Health [KIA 15508 to A.D.]. I.J. is supported by aNovartis grant (CRG) and thanks Region Ile-de-Franceand ISC-PIF for financial and logistic support. This studywas also supported by the Sixth European ResearchFramework (project number 034952, GENNETECproject), PRES UniverSud Paris, CNRS and Genopole(to F.K.). Funding for open access charge: IntramuralProgram, National Institute of Diabetes, and Digestiveand Kidney Diseases, National Institutes of Health [KIA15508 to AD].

Conflict of interest statement. None declared.

REFERENCES

1. Krivega,I. and Dean,A. (2012) Enhancer and promoterinteractions-long distance calls. Curr. Opin. Genet. Dev., 22, 1–7.

2. Ong,C.T. and Corces,V.G. (2011) Enhancer function: new insightsinto the regulation of tissue-specific gene expression. Nat. Rev.Genet, 12, 283–293.

3. Sexton,T., Bantignies,F. and Cavalli,G. (2009) Genomicinteractions: chromatin loops and gene meeting points intranscriptional regulation. Semin. Cell Dev. Biol, 20, 849–855.

4. Wallace,J.A. and Felsenfeld,G. (2007) We gather together:insulators and genome organization. Curr. Opin. Genet. Dev., 17,400–407.

5. Schoenfelder,S., Sexton,T., Chakalova,L., Cope,N.F., Horton,A.,Andrews,S., Kurukuti,S., Mitchell,J.A., Umlauf,D.,Dimitrova,D.S. et al. (2010) Preferential associations betweenco-regulated genes reveal a transcriptional interactome inerythroid cells. Nat. Genet., 42, 53–61.

6. Handoko,L., Xu,H., Li,G., Ngan,C.Y., Chew,E., Schnapp,M.,Lee,C.W., Ye,C., Ping,J.L., Mulawadi,F. et al. (2011)CTCF-mediated functional chromatin interactome in pluripotentcells. Nat. Genet., 43, 630–638.

7. Lieberman-Aiden,E., van Berkum,N.L., Williams,L., Imakaev,M.,Ragoczy,T., Telling,A., Amit,I., Lajoie,B.R., Sabo,P.J.,




ownloaded from

Dorschner,M.O. et al. (2009) Comprehensive mapping oflong-range interactions reveals folding principles of the humangenome. Science, 326, 289–293.

8. Tolhuis,B., Palstra,R.J., Splinter,E., Grosveld,F. and de Laat,W.(2002) Looping and interaction between hypersensitive sites in theactive b-globin locus. Mol. Cell, 10, 1453–1465.

9. Stamatoyannopoulos,G. (2005) Control of globin gene expressionduring development and erythroid differentiation. Exp. Hematol.,33, 259–271.

10. Palstra,R.J., Tolhuis,B., Splinter,E., Nijmeijer,R., Grosveld,F. andde Laat,W. (2003) The b-globin nuclear compartment indevelopment and erythroid differentiation. Nat. Genet., 35,190–194.

11. Vakoc,C.R., Letting,D.L., Gheldof,N., Sawado,T., Bender,M.A.,Groudine,M., Weiss,M.J., Dekker,J. and Blobel,G.A. (2005)Proximity among distant regulatory elements at the b-globin locusrequires GATA-1 and FOG-1. Mol. Cell, 17, 453–462.

12. Drissen,R., Palstra,R.J., Gillemans,N., Splinter,E., Grosveld,F.,Philipsen,S. and de Laat,W. (2004) The active spatial organizationof the b-globin locus requires the transcription factor EKLF.Genes Dev., 18, 2485–2490.

13. Song,S.-H., Hou,C. and Dean,A. (2007) A positive role for NLI/Ldb1 in long-range b-globin locus control region function.Mol. Cell, 28, 810–822.

14. Kim,S.I., Bultman,S.J., Kiefer,C.M., Dean,A. and Bresnick,E.H.(2009) BRG1 requirement for long-range interaction of a locuscontrol region with a downstream promoter. Proc. Natl Acad.Sci. USA, 106, 2259–2264.

15. Botta,M., Haider,S., Leung,I.X., Lio,P. and Mozziconacci,J.(2010) Intra- and inter-chromosomal interactions correlate withCTCF binding genome wide. Mol. Syst. Biol., 6, 426.

16. Phillips,J.E. and Corces,V.G. (2009) CTCF: master weaver of thegenome. Cell, 137, 1194–1211.

17. Murrell,A., Heeson,S. and Reik,W. (2004) Interaction betweendifferentially methylated regions partitions the imprinted genesIgf2 and H19 into parent-specific chromatin loops. Nat. Genet.,36, 889–893.

18. Kurukuti,S., Tiwari,V.K., Tavoosidana,G., Pugacheva,E.,Murrell,A., Zhao,Z., Lobanenkov,V., Reik,W. and Ohlsson,R.(2006) CTCF binding at the H19 imprinting control regionmediates maternally inherited higher-order chromatinconformation to restrict enhancer access to Igf2. Proc. Natl Acad.Sci. USA, 103, 10684–10689.

19. Farrell,C.M., Grinberg,A., Huang,S.P., Chen,D., Pichel,J.G.,Westphal,H. and Felsenfeld,G. (2000) A large upstream region isnot necessary for gene expression or hypersensitive site formationat the mouse b-globin locus. Proc. Natl Acad. Sci. USA, 97,14554–14559.

20. Bender,M.A., Reik,A., Close,J., Telling,A., Epner,E., Fiering,S.,Hardison,R. and Groudine,M. (1998) Description and targeteddeletion of 50 hypersensitive site 5 and 6 of the mouse b-globinlocus control region. Blood, 92, 4394–4403.

21. Bender,M.A., Byron,R., Ragoczy,T., Telling,A., Bulger,M. andGroudine,M. (2006) Flanking HS-62.5 and 30 HS1, and regionsupstream of the LCR, are not required for beta-globintranscription. Blood, 108, 1395–1401.

22. Splinter,E., Heath,H., Kooren,J., Palstra,R.J., Klous,P.,Grosveld,F., Galjart,N. and de Laat,W. (2006) CTCF mediateslong-range chromatin looping and local histone modification inthe beta-globin locus. Genes Dev., 20, 2349–2354.

23. Duan,Z., Andronescu,M., Schutz,K., McIlwain,S., Kim,Y.J.,Lee,C., Shendure,J., Fields,S., Blau,C.A. and Noble,W.S. (2010)A three-dimensional model of the yeast genome. Nature, 465,363–367.

24. Bau,D., Sanyal,A., Lajoie,B.R., Capriotti,E., Byron,M.,Lawrence,J.B., Dekker,J. and Marti-Renom,M.A. (2011) Thethree-dimensional folding of the alpha-globin gene domain revealsformation of chromatin globules. Nat. Struct. Mol. Biol., 1, 107–114.

25. Fraser,J., Rousseau,M., Shenker,S., Ferraiuolo,M.A.,Hayashizaki,Y., Blanchette,M. and Dostie,J. (2009) Chromatinconformation signatures of cellular differentiation. Genome Biol.,10, R37.

26. Kalhor,R., Tjong,H., Jayathilaka,N., Alber,F. and Chen,L. (2012)Genome architectures revealed by tethered chromosome

conformation capture and population-based modeling. Nat.Biotechnol., 30, 90–98.

27. Sexton,T., Yaffe,E., Kenigsberg,E., Bantignies,F., Leblanc,B.,Hoichman,M., Parrinello,H., Tanay,A. and Cavalli,G. (2012)Three-dimensional folding and functional organization principlesof the Drosophila genome. Cell, 148, 458–472.

28. Rousseau,M., Fraser,J., Ferraiuolo,M.A., Dostie,J. andBlanchette,M. (2011) Three-dimensional modeling of chromatinstructure from interaction frequency data using Markov chainMonte Carlo sampling. BMC Bioinformatics, 12, 414.

29. Mateos-Langerak,J., Bohn,M., de Leeuw,W., Giromus,O.,Manders,E.M., Verschure,P.J., Indemans,M.H., Gierman,H.J.,Heermann,D.W., van Driel,R. et al. (2009) Spatially confinedfolding of chromatin in the interphase nucleus. Proc. Natl Acad.Sci USA, 106, 3812–3817.

30. Junier,I., Martin,O. and Kepes,F. (2010) Spatial and topologicalorganization of DNA chains induced by gene co-localization.PLoS Comput. Biol., 6, e1000678.

31. Marenduzzo,D., Faro-Trindade,I. and Cook,P.R. (2007) What arethe molecular ties that maintain genomic loops? Trends Genet.,23, 126–133.

32. Dorier,J. and Stasiak,A. (2010) The role of transcriptionfactories-mediated interchromosomal contacts in the organizationof nuclear architecture. Nucleic Acids Res., 38, 7410–7421.

33. Cook,P.R. and Marenduzzo,D. (2009) Entropic organization ofinterphase chromosomes. J. Cell Biol., 186, 825–834.

34. Rosa,A. and Everaers,R. (2008) Structure and dynamics ofinterphase chromosomes. PLoS Comput. Biol., 4, e1000153.

35. Mukhopadhyay,S., Schedl,P., Studitsky,V.M. and Sengupta,A.M.(2011) Theoretical analysis of the role of chromatin interactionsin long-range action of enhancers and insulators. Proc. NatlAcad. Sci USA, 108, 19919–19924.

36. Hou,C., Dale,R. and Dean,A. (2010) Cell type specificity ofchromatin organization mediated by CTCF and cohesin. Proc.Natl Acad. Sci. USA, 107, 3651–3656.

37. Cui,Y. and Bustamante,C. (2000) Pulling a single chromatin fiberreveals the forces that maintain its higher-order structure.Proc. Natl Acad. Sci. USA, 97, 127–132.

38. Langowski,J. and Heermann,D.W. (2007) Computationalmodeling of the chromatin fiber. Semin. Cell Dev. Biol., 18,659–667.

39. Strick,T.R., Dessinges,M.-N., Charvin,G., Dekker,N.H.,Allemand,J.-H., Bensimon,D. and Coquette,V. (2003) Stretchingof macromolecules and proteins. Rep. Prog. Phys., 66, 1–45.

40. Reuillon,R., Chuffart,F., Leclaire,M., Faure,T., Dumoulin,N. andHill,D. (2010) Declarative task delegation in OpenMOLE., In:International Conference on High Performance Computing andSimulation (HPCS 2010), IEEE, Caen, June 28–July 2, pp. 55–62.

41. Kiefer,C.M., Lee,J., Hou,C., Dale,R.K., Lee,Y.T., Meier,E.R.,Miller,J.L. and Dean,A. (2011) Distinct Ldb1/NLI complexesorchestrate gamma-globin repression and reactivationthrough ETO2 in human adult erythroid cells. Blood, 118,6200–6208.

42. Dekker,J. (2006) The three ‘C’s of chromosome conformationcapture: controls, controls, controls. Nat. Methods, 3, 17–21.

43. de Gennes,P.G. (1988) Scaling Concepts in Polymer Physics.Cornell University Press, Ithaca, NY.

44. Yaffe,E. and Tanay,A. (2011) Probabilistic modeling of Hi-Ccontact maps eliminates systematic biases to characterize globalchromosomal architecture. Nat. Genet., 43, 1059–1065.

45. Hadjur,S., Williams,L.M., Ryan,N.K., Cobb,B.S., Sexton,T.,Fraser,P., Fisher,A.G. and Merkenschlager,M. (2009) Cohesinsform chromosomal cis-interactions at the developmentallyregulated IFNG locus. Nature, 460, 410–413.

46. Dostie,J., Richmond,T.A., Arnaout,R.A., Selzer,R.R., Lee,W.L.,Honan,T.A., Rubio,E.D., Krumm,A., Lamb,J., Nusbaum,C. et al.(2006) Chromosome conformation capture carbon copy (5C): amassively parallel solution for mapping interactions betweengenomic elements. Genome Res., 16, 1299–1309.

47. Simonis,M., Klous,P., Splinter,E., Moshkin,Y., Willemsen,R., deWit,E., van Steensel,B. and de Laat,W. (2006) Nuclearorganization of active and inactive chromatin domains uncoveredby chromosome conformation capture-on-chip (4C). Nat. Genet.,38, 1348–1354.




ownloaded from

48. Bau,D., Sanyal,A., Lajoie,B.R., Capriotti,E., Byron,M.,Lawrence,J.B., Dekker,J. and Marti-Renom,M.A. (2011) Thethree-dimensional folding of the alpha-globin gene domain revealsformation of chromatin globules. Nat. Struct. Mol. Biol., 18,107–114.

49. Ragoczy,T., Bender,M.A., Telling,A., Byron,R. and Groudine,M.(2006) The locus control region is required for association of themurine beta-globin locus with engaged transcription factoriesduring erythroid maturation. Genes Dev., 20, 1447–1457.

50. Song,S.H., Kim,A., Ragoczy,T., Bender,M.A., Groudine,M. andDean,A. (2010) Multiple functions of Ldb1 required for beta-globinactivation during erythroid differentiation. Blood, 116, 2356–2364.

51. Lee,H.Y., Johnson,K.D., Boyer,M.E. and Bresnick,E.H. (2011)Relocalizing genetic loci into specific subnuclear neighborhoods.J. Biol. Chem., 286, 18834–18844.

52. Hou,C., Zhao,H., Tanimoto,K. and Dean,A. (2008)CTCF-dependent enhancer-blocking by alternative chromatin loopformation. Proc. Natl Acad. Sci. USA, 105, 20398–20403.




ownloaded from

CTCF-mediated transcriptional regulation through cell type … · 2017. 1. 29. · CTCF-mediated...

Documents

Transcript of CTCF-mediated transcriptional regulation through cell type … · 2017. 1. 29. · CTCF-mediated...